Systematic Reviews: Search Protocol

1. Identify databases/resources to be searches
2. For each database
   1. Identify search terms (Term Harvesting)
   2. Develop search strategies
   3. Test-run searches, revise as needed, until a final set of search strategies is established.
3. It is often a good idea to have someone peer-review the search protocol for you before you run it.
   1. A standard checklist for Peer Review of Electronic Search Protocols (PRESS Guideline) is available from the CADTH. A 2016 update of the original 2008 guideline is available as well.

Example of a PubMed Term Harvesting Form:

• Develop search strategies specific for each database, as the strategy construction is dependent upon the database's search interface and other features.
• Don't try to have one single, perfect strategy per database. You may have one or many strategies for a single database.
• Expect to combine controlled vocabulary searching with textword searching
• Author searching can be helpful
• Be familiar with how ‘related articles” features work and track how you use them
• If you use automated limits, track how that affects the yield of your results. What are you losing?
• Record and keep ALL search strategies
• Record the yield of articles from each strategy
• Record the date on which you formally ran and collected abstracts from each search strategy

To reduce bias multiple databases must be searched. No single database will contain all of the available literature.

Each database should be searched with a database-specific search strategies

The Core databases are usually: Medline, Embase, & Cochrane Central Registry of Controlled Trials.

Common additions are CINAHL (Nursing and Allied Health literature), Science Citation Index/Web of Science. Also of note is LILACS which covers Latin American literature.

Stony Brook University has access to all of the above listed databases!

Choice of additional database depends on nature of question and may include bibliographic databases specific for Bioscience, Nursing, Pharmacology, Education, Business or other fields.

Issues include balancing sensitivity and specificity - when do you stop? Never forget that your choice of limits affects the amount of bias in your study. Access to resources may be an issue. Each database may introduce its own issues: presence/absence of controlled vocabulary, available limits and specifications, and ease or difficulty of retrieving citations.

Gathering and utilizing search vocabulary (search terms) for the literature review

• Clearly defined search question, inclusion/exclusion criteria, and outcomes required.
• A few ‘good’ articles pre-identified can be useful resources.
• Knowledge of search strategy logic and construction is valuable
• Keep EXCELLENT records (use a ‘term harvesting form’)
• Read about topic (textbooks, older reviews, other articles, other sources)
• Talk to experts (other members of team, clinicians, colleagues, etc.)
• Look at indexing of pre-identified ‘good’ articles
• Know how to use controlled vocabularies effectively and when to combine with textwords
• Know the idiosyncrasies of each database. Searching only one database is NEVER enough.
  • Not all have controlled vocabularies
  • Each controlled vocabulary tends to be unique
  • How do they handle phrase searching?
  • How do they handle limits?
  • How do they handle field searching?
  • What are the options for saving searches and exporting search results?
• Brainstorm, brainstorm, brainstorm!!!

Use available expertise ⇒ consider collaborating with your medical librarian!